Effect of gap height and non-linearity of ... - i-asem.orgEffect of gap height and non-linearity of...
-
Upload
vuongquynh -
Category
Documents
-
view
239 -
download
4
Transcript of Effect of gap height and non-linearity of ... - i-asem.orgEffect of gap height and non-linearity of...
Effect of gap height and non-linearity of incident Couette-Poiseuille flow on aerodynamic characteristics of a square cylinder near wall
*Rajesh Bhatt1), Dilip K. Maiti2) and Md. Mahbub Alam3)#
1&3)Institute for Turbulence-Noise-Vibration Interaction and Control, Shenzhen
Graduate School, Harbin Institute of Technology, Shenzhen, China 2) Department of Applied Mathematics with Oceanology and Computer Programming,
Vidyasagar University, Midnapur 721102, India #[email protected]
ABSTRACT
A numerical study on the Couette-Poiseuille based non-uniform flow over a square cylinder of height a* placed parallel to the wall at different gap ratios (L=H*/a*: 0.1≤L≤2.0, where H* is the gap between the cylinder and wall) is conducted for different inlet non-dimensional pressure gradients P. The governing equations are solved numerically through finite volume method based on SIMPLE algorithm on a staggered grid system. Both P and L have an appreciable effect on the vortex shedding frequency, aerodynamic forces and the flow structure around the cylinder. At L≤0.25 for P≤3 the flow remains steady similar to the linear shear flow (P = 0). While increasing the value at P=5, unsteadiness developed in the far field propagates towards the cylinder, becoming quasi-steady in nature for L = 0.1. A further increase in L results in stretching of the shear layers from the cylinder; for a high P the flow deflects towards the downstream side with irregular transverse and longitudinal vortices. Collecting all the simulated results obtained, four distinct flow patterns I, II, III and IV are identified based on the vorticity contours, which are steady flow, unsteady single row vortex, two stretched row of vortices and trailing edge separation with complex wake, respectively. The critical gap height for onset of the vortex shedding decreases with increasing the value of P. The intensity of peak in the power spectra of the lift coefficient increases with the increase in P. At large L, the sharp peak changes into broad band and the wake interaction becomes complicated depending on P and L. An attempt is made to find a universal shedding frequency that is independent of L for different nonlinear velocity profiles. NOMENCLATURE a* :height of the cylinder (m) L :nondimensional gap height from cylinder to plane wall P :nondimensional pressure gradient Re :Reynolds number (Ua*/ν) CD :time averaged drag coefficient CL :rms of lift coefficient St :Strouhal number U :velocity at distance h* from the wall (m/s) V : kinematic viscosity of fluid
1. INTRODUCTION In many practical cases the flow approaching the structures, for example offshore platforms, is not uniform because of the velocity gradients from the incoming flow. Although high Reynolds number flows appear in most practical examples, the flow at low Reynolds number has an application in small heat exchangers and electronic cooling components. The difference between the flow velocities at upper and bottom part of the structure in the wall proximity makes flow somewhat more complex, resulting in different wake flow patterns. Therefore, it is very important to understand the basic flow phenomena from a cylinder under the influence of a non-uniform velocity profile near the wall. This study investigates the effect of the gap height L(=H*/a*, where H is the gap between wall and the cylinder and a* is height of the cylinder) and non-linearity of approaching velocity profile based on inlet pressure gradient P on flow characteristics of a square cylinder. Numerical studies on flow past a square cylinder for different inlet shear parameter K and at different Reynolds numbers Re were conducted by Hwang et al. (1997, Re=500-1500, K=0-0.25), Cheng et al. (2005, Re=100, K=0-0.5), Cheng et al. (2007, Re=50-200, K=0-0.5), Lankadasu and Vengadesan (2008, Re<100, K=0-0.2). The results showed that the Strouhal number (St), mean drag (CD) and fluctuating forces decreases with increasing K. Cheng et al. (2007) conducted simulation at Re=100, 0≤K≤0.5 for flow over a square cylinder and concluded that unlike that for a circular cylinder, the shedding frequency for the square cylinder decreases with increasing K. Lankadasu and Vengadesan (2008) reported that with increasing K, the critical Re, at which flow becomes unsteady, is reduced. Similarly several numerical studies have been conducted on rectangular cylinder of different aspect ratios under uniform and shear flow, e.g. by Sohankar (2008, Re=105), Islam et al. (2012, Re=100-250), Yu et al. (2013, Re=105) and Cao et al.(2014, Re=22000). These studies have shown that Re effect is less significant compared to other parameters (e.g., incoming velocity profile). Cao et al. (2014) concluded that the linear movement of the stagnation point to the high velocity side with increasing K is an inherent behavior in shear flow. Several studies on effect of wall proximity of a cylinder have also been carried out (Bailey et al., 2002; Lee et al. 2005; Kumar and Vengadesan, 2007; Wang and Tan, 2008; Dhinakaran, 2011; Maiti, 2012; Samani and Bergstorm, 2015). Bailey et al. (2002) conducted an experimental study and examined the vortex shedding from a square cylinder near a wall at Re=19000 and their results showed that the suppression of vortex shedding occurs for L<0.4. Lee et al. (2005) studied effect of aspect ratio and L(0.3≤L≤2) and introduced simple passive control method to reduce the aerodynamic drag and vortex-induced oscillation. They observed that the vortex begins to shed by the interaction of the separated shear layer on the upper surface of the cylinder with the upwash flow from the gap region. Wang and Tan (2008) experimentally compared the near-wake flow patterns for a circular and a square cylinder for different L ranging from 0.1 to 1.0. They concluded that the vortex shedding strength in the case of square cylinder is relatively weaker, as compared to that of the circular cylinder at the same L, and the critical L for vortex shedding are 0.3 and 0.5 for the circular and square
cylinders, respectively. Maiti (2012) reported that aerodynamic characteristics of a square cylinder under uniform shear flow are more sensitive to L for L>1 than for L ≤1.0. Dhinkaran (2011) observed an increment in aerodynamic forces with a decrement in L for the flow past a square cylinder placed near a moving wall at Re=100. The flow behavior was classified into two-row vortex street (1≤L≤4), single-row vortex street (0.4≤L≤1), quasi-steady vortex street (L=0.3) and vortex shedding suppression (L<0.3). Recently, Samani and Bergstrom (2015) conducted large eddy simulation to examine the wall proximity effect of a square cylinder for three values of L=0, 0.5 and 1.0 at Re=500. An increase in L leads to an increase in mean lift force (CL) and a decrease in mean drag (CD) coefficients. As the cylinder approaches the wall, the wake flow becomes increasingly asymmetric with the top recirculation cell displaced significantly upward and further downstream, and a secondary recirculation zone develops on the bottom wall. The aforementioned studies on the flow around a single cylinder are based on either uniform flow or linear shear flow with and without wall proximity. It is clear from these studies that the inlet velocity profile and the gap flow greatly altered the aerodynamic forces and flow around the cylinder. However, numerical studies on a square cylinder in wall proximity under non-uniform velocity profile are scarce. Therefore, the present study is aimed at understanding the effect of inlet non-linear velocity profile on aerodynamic forces and the wake flow structure of a square cylinder at different L. 2. PROBLEM DESCRIPTION With a wall lying along the x*-axis, a cylinder of square cross section of height a* is placed parallel to the wall at gap height H* from the wall (Fig.1(a)). The inflow and outlet boundaries lie at a distance 10a* and 20a* from the front and rear face of the cylinder, respectively. The top lateral boundary lies at 10a* from the plan wall. The upstream flow field is taken as Couette-Poiseuille flow based nonlinear velocity profile u*(y*). This upstream condition is consistent with the Navier-Stokes equation and viscous effects, because of no-slip requirement at the wall (Schlichting, 2000). In other words, the cylinder is sub-merged into the boundary layer of a plane wall. The following nonlinear velocity profile (with characteristic velocity U at y* = 10a*) is considered at the inlet:
)1(1)(
*
*
*
*
*
***
h
y
h
yP
h
y
U
yu
where U is the velocity at distance h* from the wall, and P is the nondimensional pressure gradient, which is defined as
)2(2
1
22
*
*2*
dx
dpeRh
dx
dp
U
hP
where Re=Ua*/ν, p=p*/(ρU2) and, μ and ν are the viscosity and kinematic viscosity of
the fluidof P with
Fig.1.(aincident 2.1 Gov The nongiven by
. Different h fixed heig
)Schemativelocity pr
verning eq
n-dimensiny
forms of tght h=10 a
cs of flow rofiles for d
quations a
noal Navie
0 V
Vt
V
the nondimare present
configuratdifferent P,
nd numer
er-Stokes e
pV
mensional ted in Fig.
tion, (b) C, and (c) ty
rical meth
equations
Re
1 2V
velocity pr1(b).
Couette-Poypical grid d
od
for two-d
rofile u(y) f
iseuille flowdistribution
imensiona
)3(
)4(
for differen
w based nn for L=0.5
al laminar
)
)
nt values
nonlinear 5.
flow are
The nondimensional quantities V=(u, v), and p and t denote the velocity, pressure and time, respectively, with characteristic length and velocity are a* and U, respectively. No-slip condition is used on the cylinder surfaces and on the plane wall. A Dirichlet boundary (u=u(y), v=0) is used at the inlet boundary, while the Sommerfeld condition (∂ɸ/∂t)+uc(∂ɸ/∂x)=0, where ɸ is any flow variable, and uc is the local wave speed) is applied at the outlet. A slip boundary condition (∂u/∂y=0, v=0) is imposed on the top lateral boundary. The numerical treatment to the Sommerfield condition has been discussed in details in the previous study (Maiti, 2011). A finite volume method on a staggered grid system and then the pressure correction based iterative algorithm SIMPLE (Patankar, 1980) are applied. A third-order accurate QUICK (Leonard 1995) is employed to discretize the convective terms and central differencing scheme for diffusion terms. A fully implicit second-order scheme is incorporated to discretize the time derivatives. At the initial stage of motion, times step is taken to be 0.0001 which has been subsequently increased to 0.001 after the transient state. 3. GRID INDEPENDENCE AND VALIDATION OF CODE In this study, a nonuniform grid distribution is considered in the domain, with a uniform grid along the cylinder surface and increasingly enlarged grids away from the cylinder surface. The grid distribution in x-and y-directions is shown in Fig.1(c). The grid distribution in x - and y - directions is similar to that used in our previous study, Maiti (2012). Depending on the size of the distributed grids, the horizontal and vertical lines of the computational domain are divided into five and three, respectively, distinct segments. Along the y-direction, a uniform grid with size 0.005 is considered between wall and top of the cylinder (within the segment ‘Iy’ in Fig.1(c)). With reference to Franke et al. (1990) and Sohankar et al. (1997), the value of the first grid size from the cylinder is kept constant as 0.004 for the present computation. The grid refinement study on the number of uniform nodes and on the resolution of grid near the cylinder surface and the far-fields was conducted in the previous study (Maiti, 2011, 2012 and Maiti and Bhatt, 2015) for the case of single and inline tandem cylinders. A detailed discussion on the percentage of deviation has been made in these studies. In the present study the grid distribution along the y-direction varied according to the gap height of the cylinder from the wall. The numerical code validated for the case of a square cylinder placed near a wall (Bhattacharyya et al., 2006) has been used in the present study. Thus, the previous published results by the authors also show the validity of the used code for this work. In the present study the following operational dimensionless parameters affecting the flow are considered:
L= 0.1, 0.25, 0.5, 1.0, 1.5 and 2.0 Inlet pressure gradient P = 0, 1, 3, 5 Reynolds number (Re): 1000 (based on velocity at height 10a* from wall (Maiti
and Bhatt (2015))). Davis et al. (1984) found a good agreement between the numerical and experimental results for Re around 1000 for uniform flow.
Bdisrelalin
4. RESU The effediscussespectra 4.1 Flow Fig. 2 shremainsfield andSimilarlyalmost hP= 3, thinteractisteady square cFor P=5flow withthe singdominan
Fig. 2. I
Upon incdone by0.5. Thethe cylin
Bhattacharyimensiona
s assumedesults presaminar flownear flow.
UTLS AND
ect of L aned here. Tand press
w structur
hows the in steady. W
d propagaty, at L=0.2horizontallye shear layon with theflow. A mcylinder at
5, the uppeh a recircule row of c
nt peak in t
Instantane
creasing Ly Wang ande steady flonder and w
yya and Mal simulatio that the th
sented in thw for the R
D DISCUSS
nd P on flohe results ure distribu
re
nstantaneoWith increates towards25 for P ≤ y and the loyer from the counter r
more detait a lower Lr shear layulation hasclockwise vthe spectra
eous vortici
L to L=0.5, d Tan (200ow at this
wall later. A
Maiti (2004ons for diffehree-dimenhis study bRe=1000 b
SION
ow and aerare presen
ution.
ous vorticitasing P fros the cylind2 the sheaower sheahe upper srotating voled explan
L can be foyer gets sus a quasi-svortices wa of lift coe
ity contour
the flow is08) reporte
P will beAt P=1, the
4) and Merent gap hnsional effebased on thbased on
rodynamicnted and d
ty contoursom 3 to 5der, and thar layer fro
ar layer reaside of the ortices thenation on ound in Maufficient mosteady natith a defin
efficient.
rs at differe
s still steaded that the explained shear laye
aiti (2011heights anect would he assumpthe Couett
c characterdiscussed
s for L ≤ 0, unsteadinhe flow becom upper attaches thcylinder teshedding vortex sh
artinuzzi etomentum toture. Hencite peak in
ent gap hei
dy for P=0.critical L fowith the h
ers of differ
, 2012) pd Re up tonot severe
ptions of twte-Poiseuil
ristics of a in terms of
.5. At L = 0ness is decomes quaside of thee rear face
ends to roll does not o
hedding sut al. (2003o shed vor
ce the wakn the spect
ights (L ≤ 0
Previous eor onset of help of the rent streng
presented o 1400. Hoely contamwo dimensille flow ba
square cyf vorticity c
0.1, P ≤ 3,eveloped inasi-steady e cylinder e of the cy but due to
occur, geneuppression) and Maitrtices whileke is domintra, confirm
0.5) for diff
experimenf vortex she gap flow
gths from t
the two-owever, it inate the onal and
ased non
ylinder is contours,
the flow n the far in nature. emerges linder. At o weaker erating a
n from a ti (2012). e the gap nated by med by a
ferent P.
ntal study edding is between he upper
and lowlower siwake. Tone canconsequPedrizzeformatiothe cylinComparconsecustrongerstretche
Fig. 3. I The flowsmaller betweennegativeincreasesince (-deffect ofsize of nnegativearrangemincreasethat in tvortices shear lafurther istrong ifrom theand unsstrongly
er sides ofde of the
The wake in see thauent vortexetti (2001) on region nnder. Therred to the uutive vorticr coalesced vortices
Instantane
w for a largL. The vo
n the counte vortices fe in the locdP/dx) = 2f L at diffenegative ve vorticesment of ved further the lower L
are stretcayer. For ancrease innteraction e wall at lasteadiness
depends
f the cylindcylinder s
is characteat the intex formation
reported near wall srefore, theunsteady cces is enlence betwin the wak
eous vortici
ger L will bortex shedter rotatingfeatures thcal Reynol2P (1/h)2(1/erent P canortices shr, being d
vortices into 1.5 for L cases. Tched towara higher Pn L results
of vorticearger L. It t
is enhancon L and
der curl upstretches berized by aeraction tan and shethat interaurface sup
e downstrecase of L=arged at
ween the ke.
ity contour
be less infdding comg vortices dhe wake. It ds number/Re), whern be obserinks. Withdifferent in the wakeP = 1. Th
The regulards downsP the waks in the stres. There itriggers th
ced. It is cleP. With th
p into an aby the wala single ro
akes placedding occ
action of thppresses theam wake=0.25, P=5
this L. Thcylinder a
rs at differe
fluenced bmences fodue to the may be no
r that is mre h is the
erved hereh an increan intensitye. A drame flow beh
ar vortex instream. It dke is dominretched shis a pronoe instabilitear from th
he increase
alternating ll vorticity ow of vortie in the ruring behihe wall bohe positive contains , the streahe strongeand wall
ent gap hei
y the wall or L=1.0, upwash flooted that techanicall
e nondimen. For a givase in P thy and siz
matic chanhind the cyn the neardelays thenated by shear layersounced sety in the vohe Figs. 2 e in L, the
fashion. Tand dissipces. Analy
rear of thend the cyl
oundary laye vortices a
a row of amwise diser gap flowvortices,
ights (L ≥ 1
shear, comP=0. The ow from thhe increasy similar tonsional heiven P, withhe interactzes, leadsnge takes ylinder is q
wake rege roll up prstronger ps from theparation oortex distriand 3 thatfront stag
The vortex pates fasteyzing thesee cylinderinder. Zovyer and that the lowe
negative stance betwow at P=5
resulting
1.0) for diff
mpared tointeractione gap and
se in L resuo the increight from wh increasintion of poss to an
place whquite differgion destrorocess of
positive voe cylinder of the sheabution in tt the flow
gnation poi
from the er in the e figures r, with a vatto and he vortex er side of vortices.
ween the 5, results
in more
ferent P.
o that the n occurs a row of ults in an ease in P wall. The ng L, the sitive and
irregular hen L is rent from oyed and negative
ortices. A due to a ar layers the wake structure nt of the
cylinder velocity the staghigher inBased osense (Fpattern (0≤P≤3, (pattern-found aL=2); instreamw1.0≤L≤2of the cinteracti
Fig. 4. C 4.2 Gap It is cleaflow struvelocity show ththe casecases wcloser tothe jet fsteady fgap flowposition
moves toside. As r
gnation ponlet shear on the abovFig. 4). Sinare difficuL=0.25) a
-II) is founpproximate
n this rangwise direct2) and (P=3cylinder anon with wa
Classificati
p flow
ar from theucture behprofile in te mean vees where t
without vorto the lowerflow along flow. As obw for the u
of the ma
o the highereported byint, there iparameterve discussnce limitedult to asseand (P=0, d for (P=5ely when ge two rotion. Patte3, L=1.5). nd are eloall vortices
ion of flow
e discussiohind the cthe gap regelocity profhe vortex stex sheddir surface o
the wall abserved in nsteady caaximum ve
er velocityy Lankadais separatirs. sion, four dd computatess. Steady
L=0.5). Un, L=0.25),(P=5, L=0
ow of vortern-IV is In this ran
ongated incomplex w
patterns a
n made in ylinder. Thgion for diffiles for difshedding ong. Also, t
of the cylindand the veFig. 3, the
ases biaseelocity is m
y side andasu and Veion of she
distinct flowtions werey flow (pansteady pe(1≤P≤3, L
0.5), (P=3,tices are appears i
nge shear ln the transwake is for
and represe
the previoherefore, ifferent P tofferent L aoccurs, thethe positioder. For Lelocity proe flow becoed towardsmove clos
the flow engdesan ar layers f
w patterns c performed
attern-I) noeriodic flow=0.5) and , L=1.0), (observed n the ranlayers sepsverse dirrmed from
entative vo
ous sectionit is also io understand P. Lee e value of vn at which≤ 0.5 at Pfile takes
omes unstes the lowerser to the
is biased (2008), dufrom the b
can be ided, the bouoticed whew with sing(0≤P≤1, L
(0≤P≤1, L=which are
nge of aparated fromection. Duthe wake.
orticity con
n that bothnstructive nd the waket al. (200velocity is
h maximum= 0, the gaa parabolieady for L r face of thsurface of
towards tue to the sbottom face
entified in thndaries of
en (0≤P≤5gle row of
L=1.0). Patt=1.5) and e stretchepproximatem the trailue to the
tours in L-
h L and P ato investi
ke structur05) reportegreater th
m velocity oap flow is sic form ow≥ 1.0 at P
he cylinderf the cylin
he lower hifting of e for the
he broad f the flow , L=0.1), f vortices tern-III is (0≤P≤3,
d in the ely (P=5, ing edge stronger
P plane.
affect the gate the re. Fig. 5 ed that in an in the occurs is similar to
wing to a P =0. The r and the der. The
interactithe upwthe measmaller Thereforvortices of the cphenomuniform
Fig. 5. Mcylinder’ 4.3 Pres The timcylinder The PC
trend is affectednegativecylinderand it isbetweenfluctuatinmore neassociatdistributhigher nflow is oIt results
on betweeward flow nan velocityL comparere, the meshown in
cylinder anmena prese
flow and M
Mean velo’s lower fac
ssure dist
me-averagefor differe distributioattributed by increas
e for all su. As P is is more pron the surfang forces oegative PCted with thion is affec
nonlinear flobserved ds more neg
en the wakear the tra
y in the gaed to P=0.
ean velocityFig. 2 andnd a sma
ented here Maiti (2012
ocity profilece and the
ribution o
ed surfacent P at inteon around to the stesed nonlinurfaces. Tncreased, onounced aces of theon the cylinP is observhe wake vocted more ow in uppue to the n
gative pres
ke vorticityailing-edgep region p. The strony profile prd 3. Howevall region
are in a g2) for unifor
e for varioue wall. (a) P
on the cyli
e pressureermediate the surfac
ady flow aearity of thhere is a
PC is biafor larger
e cylinder nder with tved with inorticity struon the top
er surfacenegative mssure distri
y and vortie of the cyprevails annger gap frovides sover, the floof reverse
good agreerm shear fl
us L at theP = 0, (b) P
nder surfa
e distributiand large ces of the
around thehe flow at tsuction oc
ased towarP. It shouresults in
the increasncrease in ucture. It sp surface . As show
mean velocbution on t
ces attachlinder. At P
nd vortex sflow existsme insight
ow decelere flow is ement with low.
e exit posP= 5.
ace ( PC )
on ( PC )L=0.5 and
e cylinder i cylinder. the inlet. Eccurring atrds the frould be notean enhan
se in P. AloP. The dif
should be ncomparedn in Fig. 5ity near thethe rear su
hed to the P=5 (Fig. 5shedding ofor a high
t into the vrates alongobserved. those of L
ition of the
along thed 1.5 is plois not affeFor P ≥ 3,
Except at tht the front
ont top cored that thecement in ong the topfference innoted hereto other s
5, for highee bottom fa
urface (Fig
plane wal5b), an incoccurs at her P for avalidity of tg the lower
The gap Lee et al.(2
e gap betw
e peripheryotted in Figected for P, PC distrihe front fat top cornerner of thee differenctotal aver
p and botton PC distrie that the surfaces duer L and Pace of the . 6(b)).
ll causes crease in relatively
a large L. the wake r surface
velocity 2005) for
ween the
y of the g. 6(a, b). P=0. This ibution is ce, PC is er of the cylinder ce in PCrage and om faces, ibution is pressure ue to the , reverse cylinder.
Fig. 6 (surface L=0.25, in the cocore flowshear laL for L ≥the PC This indbetween
Fig. 6. T= 1.5. (cand (b). 4.4 Time Contourcylinder differententire LP≥1, witfor all L & II, flowto L. Thedistributdifferencsmaller Howeve≈ 1.5, thcomparecylinder large prcylinderL=2. Co
(c) shows of the cylithat the dore of the w resemb
ayers of the≥ 0.5. At L=on the wa
dicates than the uppe
Time-averac)Pressure
e average
rs for the tion the L-
t flow patte-P plane ath change and secon
w is either e directionion on thece betweenCD. For se
er, this incrhe flow paed to the lat large L
ressure di. For L ≤ 0
orrespondin
PC distribnder and aifference bgap regio
les that ofe cylinder i=0.5 the stall is alwayat the strear and lowe
aged presse distributio
ed drag co
ime averag-P plane aerns are aas expectein L. The v
nd: P≥1 forsteady or
n of the drae surface on the front
econd regimement for
ast cylinderower L. Th
L. As can bifference w
0.5, CD for ng value o
bution in talong the pbetween Con. This imf a channeis lost (Figtagnation ps higher thamline cur
er shear lay
sure coefficon in gap f
oefficient (
ged drag care shownlso marke
ed. In genvariation ofr L>0.5. Fofluctuation
ag force onof the cylinand rear f
me (patternP=1 case r will be inherefore, itbe seen frwhich conP=0 is 0.0
of CD for P
the gap replane wall,
PC at the cmplies that el flow. As. 2). The Cpoint is shihan that alrvature isyers of the
cient ( PC ) for differen
(CD) and fl
coefficient n in Fig.7. d in the fig
neral, CD sf CD consis
or first two n is very len the cylindnder (Fig. faces of then-III & IV),is less. Wh
nfluenced lt displays rom Fig. 6 ntributes th022, and thP=5 are 0.
egion of th, at differecylinder suthe gap fl
s a result, PC distribu
ifted towarong the bopositive, acylinder (F
on the cylint L for P =
uctuating
(CD) and rThe boun
gure. The shows quasts two regregimes co
ess and theder was als6). At lowe cylinder CD increahen the cyless by thelarger varifor higher
he higher he value sli47 and 3.
he cylindent L for a rface and ow is unidthe intera
ution changrds the lowottom surfaand there Fig. 3).
nder surfa= 3. Legen
lift coeffi
rms lift coendaries cor
CD is alwalitatively simes: L≤0
orrespondie change iso confirme
wer gap heis very lesse gradualinder is ple plane waiation of for P at a pa
total aveightly incre85, respec
r, along thparticular wall is alm
directional,action betwges with c
wer side. Foace of the is strong
ace: (a) L=0nds are sam
cient (CL)
efficient (Crrespondinays positivsimilar beh.5 for P≥1 ing to flow in CD is ined by the
eights the ss (Fig. 6), ally with gaaced at thall shear l
orces actinarticular L,
erage drageased up toctively. The
he lower P=3. For
most zero and the
ween the hange in or L=1.5, cylinder. coupling
0.5, (b) L me in (a)
)
CL) of the ng to the ve on the havior for
and P=0 pattern-I sensitive pressure pressure resulting p height. e large L ayers as
ng on the , there is g on the o 0.12 at e rate of
increme
Fig. 7. C(C’L) at d The varlower fluIII and IVto the Cfluctuatiovortex sfluctuatioapproximstructurehigher vthat the stagnatiCD and C 4.5 Vort The speFig. 8. Fand theris obserthe peakwall vorbroadbafrequenc
nt in CD de
Contours odifferent L
riation of Ructuation reV). CL is re
CD coefficieon at loweshedding on of lift mately 100es in Figs.velocity reg
variation ion point oCL values
tex shedd
ectra of lift For L ≤ 0.1re was no rved in thek height bortex interaanded peacy and the
epends on
of (a) time aand P.
RMS lift coegime (for elatively lo
ent, CL attrer L and Pregime fois stronge
0% for P=52 and 3, wgion. Lei ein the forc
of the cylinwith increa
ing freque
coefficien, P ≤ 5 anpeak the p
e spectra. Foosting witaction resuks. These interaction
P and it in
averaged d
oefficient (pattern I a
ow for P=0 ributes eithP. Quick inr all L an
er. At L=2 5 case com
with the incet al. (199e coefficieder. Maiti ase in L fo
ency and s
ts of the cnd L = 0.25power specFor P = 0, th increasinults in a c
multiples n of the wa
ncreases m
drag coeffi
(CL) in L-Pand II), andcase com
her supprencreasing nd P. Hig
the fluctumpared tocrease in L99) and Dients is asso
(2012) obor the unifo
spectra
cylinder as5, P ≤ 3 anctra. For hione single
ng L. For hcomplicatepeaks cor
ake with wa
monotonica
cients (CD)
P plane ald quickly inpared to th
ession of vregime is her fluctuuation of t P=0. As
L the stagnpankar anociated wit
bserved theorm shear f
functions nd L = 0.5igher valuee dominanhigher valued wake (rrespond toall vortices
ally with P a
) and (b) rm
lso consistncreasing he higher vvortex shed
corresponation at hthe lift coeit can be s
nation poind Sengupth the move similar trflow.
of L and P, P = 0, thes of L andnt peak is iues of P anFig. 3) ano the cylin
s.
at a particu
ms lift coef
ts of two regime (fo
values of Pdding or vends to the higher P efficient eseen from
nt is shiftedpta (2005) vement of rend of inc
P are prese flow wasd P, a definidentified (nd L, the wnd multiplender own s
ular L.
fficient
regimes: or pattern P. Similar ery small onset of indicates nhanced the flow
d into the reported the front
crease in
sented in s steady, nite peak (Fig. 8a), wake and e and/or shedding
F
Fig. 9 prcylinder considerplaced increaseis contra
Fig. 8. The
Fig.
resents thefor differe
red as the in nonuni
es when P ary to the
power spe
9. Contour
e contour ent L and St. As demform flow
P is further finding of
ectra of flu
rs of Strou
plots of StP. The do
monstratedcrucially
increasedSaha et a
uctuating lif
hal numbe
trouhal numominant ped in Figs. 2depends from 0 to
al. (1999)
ft coefficien
er (St) at di
mber (St) oeak in Fas2 & 3, the w
on P an5. The incand Lanka
nt at differe
fferent L a
of the vortst Fourier wake topold L. The
crease in Sadasu & V
ent P and L
and P.
tex sheddintransform logy of the
Strouhal St for preseVengadesa
L.
ng of the (FFT) is
e cylinder number
ent study n (2008)
who reported that St either remain constant or decreases with increase in the incoming shear rates. The maximum St (=0.18) is identified in pattern-II for L=0.25, P=5, where the wake consists of a single row of regular vortices. St of the dominant peak is sensitive to P. In general for pattern-III and IV, St decreases with increase in L for P ≥ 1. At large L, the effective Reynolds number increases and vortices are stretched, it results smaller value of St. These observations are consistent with Maiti (2012) who reported drop in St at higher L for a uniform shear flow past square cylinder.
It has been observed here that significant changes occur in St with change in L particularly for higher values of P. For P≤3 variation in St is less with variation in L, compared to the higher value of P ≥ 3. The variation in St is corresponding to change in wake flow pattern. Roshko (1954) define a universal Strouhal number (St*) based on the hypothesis that the wake of different bluff bodies are similar in structure, as
)5(*
*
a
w
u
USt
u
wfSt
bb
Where f is the shedding frequency, w is the lateral distance between the two free shear layers and is the time mean velocity at the boundary layer separation defined as
)6(1 P
bC
U
u
Where CP is the base pressure. The concept of a universal Strouhal number has been used by many researchers with different definition of wake widths (Griffin & Ramberg 1974, Roshko 1954, Alam & Zhou 2007 and Alam et al. 2011). All these studies are based on higher Reynolds number uniform flow. So far, nobody has used this theory for the single cylinder with nonuniform inlet flow in wall proximity. An interesting question arises is : for a nonlinear inlet flow, is it possible to find the universal Strouhal number that is independent of L ? It is clear from the flow structures that the presence of incoming nonlinear flow and the wall confinement produce an asymmetry in the two separated shear layers emerging from the upper and lower surfaces of the cylinder. Therefore, it is difficult to measure the wake width for the present case. For the St* calculation we have considered cylinder height (a*) as the characteristic length instead
of the wake width w. The values of St* are calculated as, buStSt /* and plotted
along with St in Fig. 10. It is clear from Fig.10, St and St* curves are qualitatively similar. For linear shear flow case (P=0), the values of St* collapsed to fairly constant value with less variation (0.037±0.003) with L. However, there is large variation in St* for P=5 (St*=0.085 ± 0.03). The higher variation in St* for larger P can be explained based on that increase in P is similar to increase in Re. Richter and Naudascher (1976) observed variation in St* (0.1560-0.2800) with change in Re from 2×104 -1.0×105 to 2×105-4×105. Here also for P≥3, we observe large variation in St* ≈ 0.03 - 0.1 for P=3 and St* ≈ 0.05-0.13, for P = 5 with L. However, the range of variation in St* is (0.036 - 0.04) & (0.04 - 0.058) for P = 0 & 1, respectively. Large variation at lower L(≤1.0) is attributed the wall proximity effect, which insinuates that the defining of universal Strouhal number St*
does noFig. 10 tthat RosP for L≥
Fig. 10. for differ 5. CONC The effea squargradientflow strubased oP=0, L=it conveincomingin patterand two is obserchange large L becomepower sdependsfor P ≥
ot fit for thethat for L≥shko theor1.0.
Variationsrent P.
CLUSIONS
ect of gap hre cylindert (P) and gucture, forcon the cont0.5 in patte
erts into cog flow at Prn-II. In parow of vor
rved from tin either Pand P in s complic
spectra. Ths on P. The3 compar
e flow past1.0, variat
ry for findin
s in Strouh
S
height andr has beegap height ces and shtours of spern-I. The
omplete voP=5. Unstettern-III shrtices stretthe cylinde
P or L whicpattern-IV,ated, whic
he critical ge sheddingred to P=0
t square cyion in St* ing St* bas
hal numbe
d inlet-flow en investig
(L) have hedding frepanwise vounsteadine
ortex sheddeady periodhear layerstched in theer for all thch play a m, trailing ech resultsgap heightg frequenc0 case. Th
ylinder in wis less par
sed on
r (St) and
nonlinearigated numan apprec
equency. Forticity. Thess is geneding at L=dic flow wi
s from the e streamw
he cases. Tmeasure rodge separStrouhal
t for onset cy of the cyhe St* va
wall proximrticularly foprovides b
modified S
ty on the cmerically. Wciable effecFour distince flow is serated for
=0.25 by inith a singlebottom of
wise directioThe gap floole in the foration occupeak becor suppre
ylinder decriation sho
mity. As it or small vabetter appr
Strouhal n
characterisWe found ct on the dct flow pattsteady for L=0.1 at fa
ntroducing e row of vothe cylindeon. For L ≥ow changeormation ours and thcomes brossion of th
creases witows that R
is evident alues of P. roximation
number (St
stics of flowthat the
dependencterns are cP≤ 3, L≤ 0
ar downstrnonlineari
ortices is oer remain ≥ 1.0 unstees appreciaof wake voe wake in
oad bandehe vortex sth increaseRoshko’s t
from the It shows at lower
t*) with L
w around pressure
ce on the classified 0.25 and eam and ity in the observed attached
eady flow ably with rtices. At teraction d in the shedding e in the L theory of
universal Strouahl number does not fit for higher P≥3 and lower L≤1.0. The values of CD and CL are insensitive to change in L for P=0 and upto L=0.5 for P ≥ 1. For L>0.5, CD increases almost linearly with L. The average pressure coefficient PC shows significant change along the surface of the cylinder with increase in L and P that result in an increase in the value of CD. ACKNOWLEDGMENTS Alam wishes to acknowledge the support given to them from National Natural Science Foundation of China through Grants 11672096 and from Research Grant Council of Shenzhen Government through grant JCYJ20160531191442288. REFERENCES Alam, M.M. and Zhou, Y. (2007), “Turbulent wake of an inclined cylinder with water running”, J. Fluid Mech. 589, 261-303. Alam, M.M., Zhou, Y. and Wang, X.W., (2011), “The wake of two side-by-side square cylinders”, J. Fluid Mech. 669, 432-471. Bhattacharyya, S. and Maiti, D.K. (2004), “Shear flow past a square cylinder near a wall”, Int. J. Eng. Sci., 42, 2119-2134. Bhattacharyya, S., Maiti, D.K. and Dhinakaran, S., 2006, “Influence of buoyancy on vortex shedding and heat transfer from a square cylinder in wall proximity”, Numer. Heat Transfer, Part A, 50(6), 585–606. Bailey,S.C.C., Kopp, G.A. and Martinuzzi, R.J. (2002), “Vortex shedding from a square cylinder near a wall”, J. Turbulence, 3, 003. Cao, S., Zhou, Q. and Zhou, Z. (2014), “Velocity shear flow over rectangular cylinders with different side ratios”, Computers & Fluids, 96, 35-46. Cheng, M., Whyte, D.S. and Lou, J. (2007), “Numerical simulation of flow around a square cylinder in uniform-shear flow”, J. Fluids Struct., 23, 207-226. Cheng, M., Tan, S.H.N. and Hung, K.C. (2005), “Linear shear flow over a square cylinder at low Reynolds number”, Physics of fluids, 17, 078103. Dhinakaran, S. (2011), “Heat transport from a bluff body near a moving wall at Re= 100”, Int. J. Heat Mass Transfer, 54, 5444-5458. Dipankar, A. and Sengupta, T.K. (2005), “Flow past a circular cylinder in the vicinity of a plane wall”, J. Fluids Struct., 20, 403–423. Davis, R.W., Moore, E.F. and Purtell, L.P. (1984), “A numerical-experimental study of confined flow around rectangular cylinders”, Phys. Fluids, 27, 46–59. Franke, R., Rodi, W. and Schonung, B., (1990), “Numerical calculation of laminar vortex shedding flow past cylinders”, J. Wind Eng. Aerodynamics,35(1–3), 237– 257. Griffin, O.M. and Ramberg, S.E. (1974), “The vortex street wakes of a vibrating cylinders”, J. Fluid Mech. 66, 729 -738. Hwang, R.R. and Sue, Y.C. (1997), “Numerical simulation of shear effect on vortex shedding behind a square cylinder”, Int. J. Num. Meth. Fluids, 25, 1409-1420. Islam, S.U., Zhou, C.Y., Shah, A. and Xie, P. (2012), “Numerical simulation of flow past rectangular cylinders with different aspect ratios using the incompressible lattice
Boltzmann method”, J. Mech. Sci Tech., 26, 1027-1041. Kumaran, M. and Vengadesan, S. (2007), “Flow characteristics behind rectangular cylinder placed near a wall”, Numerical Heat Transfer-A, 52, 643-660. Lankadasu, A. and Vengadesan, S. (2008), “Onset of vortex shedding in planar shear flow past a square cylinder”, Int. J. Heat Fluid Flow, 29, 1054-1059. Leonard, B.P. (1995), “Order of accuracy of QUICK and related convection–diffusion Schemes”, Appl. Math. Model. 19, 640–653. Lee, B.S., Kim, T.Y. and Lee, D.H. (2005), “Control of vortex shedding behind a rectangular cylinder near the ground”, Numerical Heat Transfer-A, 47, 787-804. Lei, C., Cheng, L. and Kavanagh, K. (1999), “Re-examination of the effect of a plane boundary on force and vortex shedding of a circular cylinder”, J. Wind Eng. Aerodynamics, 80, 263–286. Maiti, D.K. (2011), “Dependence of flow characteristics of rectangular cylinders near a wall on the incident velocity”, Acta Mech, 222, 273-286. Maiti, D.K. (2012), “Numerical study on aerodynamic characteristics of rectangular cylinders near a wall”, Ocean Eng., 54, 251-260. Maiti, D.K. and Bhatt, R. (2015), “Interactions of vortices of a square cylinder and a rectangular vortex generator under Couette Poiseuille flow. ASME J Fluids Eng; 137, 051203-1. Martinuzzi, R.J., Bailey, S.C.C.R. and Kopp, G.A. (2003), “Influence of wall proximity on vortex shedding from a square cylinder”, Exp. Fluids, 34, 585–596. Patankar, S.V. (1980), “Numerical Heat Transfer and Fluid Flow”. Hemisphere Publishing Corporation, Taylor & Francis Group, New York. Roshko, A. (1954), “On the drag and shedding frequency of two dimensional bluff bodies”, NACA Tech. Note no. 3169. Richter, A. and Naudascher, E. (1976), “Fluctuating forces on a rigid circular cylinder in confined flow”, J. Fluid Mech., 78(03), 561–576. Samani, M. and Bergstrom, D.J. (2015), “Effect of a wall on the wake dynamics of an infinite square cylinder”, Int. J. Heat Fluid Flow, 55, 158-166. Schlichting, H. and Gersten, K. (2000), “Boundary-Layer Theory”, Springer, New York. Sohankar, A., Norberg, C. and Davidson, L. (1997), “Numerical simulation of unsteady low-Reynolds number flow around rectangular cylinders at incidence”, J. Wind Eng. Aerodynamics, 69,189–201. Sohankar, A. (2008), “Large eddy simulation of flow past rectangular-section cylinders: side ratio effects”, J. Wind Eng. Aerodynamics, 96, 640-655. Saha, A.K., Biswas, G., and Muralidhar, K. (1999), “Influence of inlet shear on structure of wake behind a square cylinder”, Journal of Engineering Mechanics, 125(3), 359- 363. Wang, X.K. and Tan, S.K. (2008), “Comparison of flow patterns in the near wake of a circular cylinder and a square cylinder placed near a plane wall”, Ocean Eng. 35, 458-472. Yu, D., Butler, K., Kareem, A., Glimm, J. and Sun, J. (2013), “Simulation of the Influence of aspect ratio on the aerodynamics of rectangular prisms”, J. Eng. Mechanics ASCE, 139, 0733-9399. Zovatto, L. and Pedrizzetti, G., (2001), “Flow about a circular cylinder between parallel walls”, J. Fluid Mech., 440, 1 - 25.